Open-source game analytics for studios

Game analytics your whole studio can use and own

FOR THE WHOLE TEAM

Dashboards with your core metrics: retention, level funnels, session length

AI Agents layer
Opencode
OpenClaw
Codex
Claude

FOR EVERYONE

Just ask. Your AI agent is connected to the database. Get answers in minutes, not hours

Where do players fail most?
level_idfailuresplayers
124,8123,204
73,9772,911
182,5631,847
41,9211,540
231,4061,102

FOR ANALYSTS

Go as deep as you want querying data with SQL

Rawbbit collects every event from your game, stores it in a database you own, and helps your team find answers through SQL, dashboards, or AI agents connected directly to your data.

Open source. No subscriptions. No per-user fees. No per-event pricing.

Ask questions about your game data in plain language

Most studios don’t need more charts. They need simple answers about what players do in their game. Rawbbit helps producers, game designers, product analysts explore player data themselves and get answers faster with AI Agents

Self-host it free

What's included in your game analytics stack

Rawbbit is one tool that covers the whole path — from a player tapping a button to your team having an answer.

Everything your game does, collected

Every action players take — levels started and failed, purchases, sessions, ad views, etc. — is collected automatically and kept in full detail, not aggregated away.

Event pipeline · handles launch-day traffic spikes

Database you own, not someone else's

All of it lands in a database that belongs to you. Nothing is locked behind a vendor's export limits, and nothing gets deleted because a plan changed.

Self-hosted or managed · export everything, any time

Dashboards the whole team reads

Retention, revenue, level progression, funnels, cohorts — set up and ready to look at, so everyone works from the same numbers.

Metabase · built around your game, not a generic template

AI agent your team can just ask

Ask a question the way you'd ask a colleague. The AI agent queries your own database and answers — no SQL, no waiting. Give it access to your code base too, and it connects events to the levels and systems they came from.

Built into the stack through the MCP server — not a chatbot bolted on top

See why players fail a level, not just that they do

Game analytics tells you that level 12 completion dropped. It rarely tells you why: which move limit, which blocker, which change in the last build.

WHAT YOUR ANALYTICS SHOWS

Level 12 completion rate dropped 18% after v1.4

+ WITH CODE BASE ACCESS

v1.4 cut level 12 to 22 moves and added an ice blocker that needs 3 matches to clear

Give the AI agent read access to your code base and it links player behavior to the levels, systems, and mechanics that produced it. For example in a match-3 game that means connecting a drop in level completion rate to the move limit you lowered and the blocker you added. The root cause, not just the metric.

Read-only, and scoped to what you choose. The AI agent can be pointed at a single system or level pack rather than your whole project. Code base access is optional: it works on your analytics data alone until you turn it on.

Use the AI agent your team already has. Rawbbit connects to Claude, Codex, OpenClaw or Opencode through an MCP server, so there's nothing extra to buy and no new vendor between your team and your code.

Want the technical details, or want to talk it through?

How it works

Self-hosted game analytics, free or fully set up

The same open-source stack, two ways to start.
No subscriptions, no per-event pricing, no per-seat fees.

Self-host it free

€0

The whole thing is open source. Deploy it yourself on your own infrastructure and run it however you like.

Self-host it free

Running a small studio? We're open to shared database access arrangements.
Ask us at team@rawbbit.one

Own your game analytics data: open source, portable, no lock-in

Rawbbit is open source under Apache 2.0. Events arrive over HTTP, NATS JetStream buffers the write path, and a raw writer lands partitioned Parquet files in S3-compatible object storage you own. From there the raw layer loads into ClickHouse, and your team reads it through Metabase dashboards, direct SQL, or AI agents. Every piece is open source. You can stand the whole stack up locally with Docker Compose, and two VMs are enough to run it in production: one for ingestion and raw storage, one for ClickHouse and the access surfaces.

Raw Parquet is the system of record, which matters most on the day you want to leave. Your full event history is already in an open format, so you point another warehouse at the same files and keep going. Nothing needs re-tracking, and nothing waits on an export request.

The Rawbbit MCP server exposes a read-only analytical surface over your ClickHouse events table, so Codex, OpenCode, OpenClaw or any other MCP client can explore your data without touching the ingestion runtime. Container logs are reachable the same way, which means the AI agent that answers your analytics questions can also help you work out why events stopped arriving.

No subscriptions. No per-user fees. No per-event pricing. A practical open-source alternative to Firebase Analytics, Amplitude, devtodev, and similar tools.

Who Rawbbit is for

A good fit

  • Game studios that want real analytics without hiring a data engineer
  • Teams where designers and producers need answers fast, not ticket queues
  • Studios that care about owning their player data long-term
  • Teams outgrowing free tiers or facing usage-based pricing they can't predict

Not a fit

  • Teams wanting a zero-setup plug-and-play dashboard product
  • Studios where nobody will ever look at the data
  • Teams unwilling to own data
Founder of Rawbbit

Why I built Rawbbit

Built from a real game analytics cost problem

I work in data engineering and have spent time around game development teams where analytics volume grows fast and platform limits become real. I've seen what happens when Firebase Analytics export to BigQuery hits the 1 million events per day limit: either you pay for the enterprise version or you start losing data.

Read more about the founder →

Frequently asked questions

Not to run it. We deploy and manage the pipeline and the database, dashboards come ready, and the AI agent answers questions in plain language. But someone at the studio still needs to look at the numbers and decide what to do with them. Rawbbit removes the infrastructure work, not the thinking. If you do have technical people, they get full SQL access to the database.

Apache 2.0, all source code public on GitHub. You can read it, self-host it, fork it, or take your data and leave at any point.

Yes. Self-hosting is free and fully supported. We can help to set up Rawbbit on your or our managed servers and provide maintain services. Contact us if you need help: team@rawbbit.one

In object storage you control, in an open file format, plus a ClickHouse database for fast queries. Both live in infrastructure you own unless you specifically ask us to host it.

Costs stay tied to the infrastructure you run, not to how many events you send. There's no per-event pricing, so a successful launch doesn't turn into a surprise bill.

Yes. Raw events stay in an open, portable format specifically so this is possible.

It's built for studios moving away from the tools where usually they hit event limits, pricing at scale, or not being able to get at their own raw data without additional costs.

Endpoint is already available to get event data from your client or backend. Native game engine SDKs are in progress — talk to us about your setup and timeline.

Only if you give it access. With code base access it can connect analytics events to the specific game mechanics and systems behind them. Without it, the AI agent works on your analytics data alone.

See what your players are actually doing

Book a demo to see all Rawbbit features. We'll explain how you can run game analytics on your own servers — event tracking pipeline, database, dashboards, and AI agent included.